Binaural and cochlear disparitiesPhilip X. Joris, Bram Van de Sande, Dries H. Louage, and Marcel van der Heijden doi:10.1073/pnas.0601396103 2006;103;12917-12922; originally published online Aug 14, 2006; PNAS This information is current as of March 2007. & ServicesOnline Information www.pnas.org/cgi/content/full/103/34/12917etc., can be found at: High-resolution figures, a citation map, links to PubMed and Google Scholar, References www.pnas.org/cgi/content/full/103/34/12917#BIBLThis article cites 23 articles, 11 of which you can access for free at: www.pnas.org/cgi/content/full/103/34/12917#otherarticlesThis article has been cited by other articles: E-mail Alerts. click hereat the top right corner of the article orReceive free email alerts when new articles cite this article - sign up in the box Rights & Permissions www.pnas.org/misc/rightperm.shtmlTo reproduce this article in part (figures, tables) or in entirety, see: Reprints www.pnas.org/misc/reprints.shtmlTo order reprints, see: Notes:Binaural and cochlear disparitiesPhilip X. Joris*†‡, Bram Van de Sande*, Dries H. Louage*, and Marcel van der Heijden**Laboratory of Auditory Neurophysiology, University of Leuven, Campus Gasthuisberg, B-3000 Leuven, Belgium; and†Department of Physiology, University of Wisconsin Medical School, 1300 University Avenue, Madison, WI 53706Edited by Eric I. Knudsen, Stanford University School of Medicine, Stanford, CA, and approved June 27, 2006 (received for review February 18, 2006)Binaural auditory neurons exhibit ‘‘best delays’’ (BDs): They aremaximally activated at certain acoustic delays between sounds atthe two ears and thereby signal spatial sound location. BDs arisefrom delays internal to the auditory system, but their source iscontroversial. According to the classic Jeffress model, they reflectpure time delays generated by differences in axonal length be-tween the inputs from the two ears to binaural neurons. However,a relationship has been reported between BDs and the frequencyto which binaural neurons are most sensitive (the characteristicfrequency), and this relationship is not predicted by the Jeffressmodel. An alternative hypothesis proposes that binaural neuronsderive their input from slightly different places along the twocochleas, which induces BDs by virtue of the slowness of thecochlear traveling wave. To test this hypothesis, we performed acoincidence analysis on spiketrains of pairs of auditory nerve fibersoriginating from different cochlear locations. In effect, this analysismimics the processing of phase-locked inputs from each ear bybinaural neurons. We find that auditory nerve fibers that innervatedifferent cochlear sites show a maximum number of coincidenceswhen they are delayed relative to each other, and that theoptimum delays decrease with characteristic frequency as in bin-aural neurons. These findings suggest that cochlear disparitiesmake an important contribution to the internal delays observed inbinaural neurons.auditory nerve 兩 phase-locking 兩 sound localization 兩 stereo 兩temporal codingAremark able feat of the human auditory system is itsextraordinary sensitivit y to differences in the ac oustic wave-for ms between the two ears. Interaural time differences (ITDs)arise when sound sources are offset from the midline toward oneside of the head, and differences in ITD can be perceived tovalues of 10–20␮s. Most computational models of this ability arebased on the qualitative scheme of Jef fress (1) in which apopulation of binaural cells effectively cross-correlates the twomonaural signals by a process of coincidence detection and delaylines. In this scheme, the input to a binaural neuron from one earis delayed relative to that of the other ear by a so-called ‘‘internaldelay,’’ generated by differences in length of the axonal pathwaysf rom each side. The binaural neurons are coincidence detectorswhose inputs are spiketrains that are time-locked to the ongoingfeatures of the sound stimulus at each ear. Only spiketrains thatarrive coincidentally activate the binaural cell: This activation isachieved at an ITD (‘‘best delay’’ or BD) that exactly compen-sates for the internal delay. By arranging these axons in adelay-line pattern, a range of internal delays to different coin-cidence detectors is created.There is physiological and anatomical support for the Jeffressmodel both in the medial superior olive (MSO) of mammals andin nucleus laminaris of the barn owl (2, 3). However, binauraldat a in the mammalian inferior colliculus (IC), which receivesinput from the MSO, show a systematic relationship bet weenBDs and the f requency to which neurons are most sensitive(characteristic frequency or CF). BDs are, on average, small athigh CFs and large at low CFs (4, 5). To fit the Jeffress model,these findings would imply longer axonal delay lines at low CFsthan at high CFs.Schroeder (6) suggested an alternative scheme, which we refer toas the cochlear disparity hypothesis. Sounds cause a propagatingwave in the cochlea, so that high-frequency components are trans-duced earliest at the base of the cochlea, and low-frequencycomponents are transduced several milliseconds later at its apex. Ifthe monaural inputs to a binaural neuron were derived from slightlydifferent places in the left cochlea vs. the right cochlea, an internaldelay would result. Indeed, computational models (7, 8) suggestthat small cochlear mismatche s should have significant effects onBD. The only experimental test to date, in the barn owl, concludedthat there was no need to invoke nonaxonal delays (9). Althoughthere have been no experimental studies in the mammal thatspecifically address the cochlear disparity hypothesis, some featuresin binaural response s have been interpreted as indicative of suchmismatches (10).The effect of cochlear disparit y is more complex than a simpletime delay because it reflects differences in both the magnitudeand phase spectra of two points in the c ochlea. The place tomeasure these effects in pure form is not in the binaural systembut rather in the output of the cochlea: the auditory nerve. Here,we use a coincidence analysis to measure the magnitude of theef fective delays resulting from cochlear disparities, as a functionof CF. We processed monaural (auditory nerve) spike trains,obt ained within a single ear, through the simplest conceivablec oincidence detector, and we c ompared this output with binaural(IC) responses obt ained with